1 / 19

Advanced Computer Graphics Advanced Shaders

Advanced Computer Graphics Advanced Shaders. CO2409 Computer Graphics Week 16. Lecture Contents. Introduction Per-Pixel Lighting Normal Mapping Parallax Mapping Cell Shading Other Techniques Further Reference. Introduction.

cturley
Télécharger la présentation

Advanced Computer Graphics Advanced Shaders

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Advanced Computer GraphicsAdvanced Shaders CO2409 Computer Graphics Week 16

  2. Lecture Contents • Introduction • Per-Pixel Lighting • Normal Mapping • Parallax Mapping • Cell Shading • Other Techniques • Further Reference

  3. Introduction • We have looked at the basics of the entire pipeline, using vertex and pixel shaders • We have also created some interesting effects using shader techniques • But these effects were of limited use [wiggle!] • We can use advanced shaders to implement more useful high-quality rendering techniques • Modern applications & games make much use of such shaders • Here we briefly introduce some key techniques • Not in depth, intended to give a flavour only • Overview of techniques is examinable, exact knowledge of the finest detail is not expected

  4. Per-Pixel Lighting • We have considered lighting as a vertex processing task • Vertex lighting colours were interpolated (blended) across polygon pixels • However, there are problems lighting large polygons only using the vertices • Specular highlights can be lost or vague • Attenuation (fall off of brightness) is poorly represented • We can move the lighting calculations to the pixel shader • Very precise lighting effects are possible • More expensive – many more lighting calculations

  5. An Aside: Rasterisation • Vertex shaders work on vertices, pixel shaders on pixels • Many pixels per triangle, how are they generated? • The rasterization stage, occuring between the shaders converts triangles into pixels • Take three vertices forming a triangle, calculate every pixel inside • Call pixel shader once for every pixel • Data from vertices is interpolated to give input to pixel shader • Clearly pixel shader performance is important

  6. An Aside: Interpolation • The vertex shader outputs data into the pipeline • E.g. colours, texture coordinates, etc. • Geometry shader may update this data, but we’ll ignore that here • This data is linearly interpolated by the rasterization stage to give inputs to the pixel shader • Each pixel is given a weighted average of the outputs from the three vertices in the triangle • In fact some perspective correction also occurs • Some data doesn’t interpolate ideally • E.g. Vectors change length when interpolated

  7. Per-Pixel Lighting • Per-pixel lighting: • Note the reflection on the front of the cube • Doesn’t affect any vertex • Vertex processing would lose this highlight entirely • Also attentuation & specular on the floor would be lost Method: • The vertex shader passes the vertex position and normal on to the pixel shader • Interpolated & used by the pixel shader to calculate lighting with the usual equations • I.e. Just pass data down the pipeline and calculate lighting later

  8. Specular / Gloss Maps • Multiple textures can be blended on a polygon • Trivial in a shader • E.g. a specular map (or gloss map) is a texture used to adjust specular highlights • It defines which areas are affected by specular light • i.e. the shiny areas

  9. Diffuse & Specular Maps • Equation used in pixel shader to use a specular map: Output pixel colour = diffuse light * diffuse map + specular light * specular map • The main texture is called the diffuse map • The textures are named by the lighting they affect • Convenient to store the diffuse and specular map within a single bitmap • Diffuse in the r, g, & b channels, specular in the alpha channel • Can’t use alpha transparency + =

  10. Normal Mapping • When calculating lighting per-pixel, the normals are interpolated: • All polygons will be flat or smoothly curved • Can only represent bumpiness with detailed models • Use yet another texture containing normals – a normal map • Note the use of multiple textures in more advanced effects • Look up the normal for each pixel in this texture • Ignore interpolated vertex normals • Improves quality greatly • But some technical complexity

  11. Normal Mapping • Normal mapping example: • Diffuse texture is very simple • Normal map shown later • Normal mapping suggests bumpiness when combined with pixel lighting

  12. Normal Map Example • Normal maps look unusual • Look at red, green and blue channels to visualise + + Red X direction of normal = Green Y direction of normal Actual Normal Map Blue Z direction of normal

  13. Normal Map Transformation • Normals in a normal map are defined relative to the polygon texture they are applied to • Assumes texture is facing down the Z-axis • This is tangent space – the world as seen from the texture • But textures in the scene might face in any direction • So transform normals from normal map onto mesh • This is a matrix transform: • Tangent space (relative to texture) • To world space (because lights are in world space)

  14. Tangent Space • We need to calculate a matrix to represent the viewpoint of each polygon - the tangent space of the polygon • X & Y axes match U & V texture axes • The Z axis is simply the polygon normal • Already have the polygon normal, • Store tangent vector (texture U axis) • But calculate the bi-tangent (see lab) • Normal maps use RGB to define (X,Y,Z) vectors in tangent space • RGB range is 0-1, vector ranges -1 to 1 • So conversion: (RGB values * 2) – 1 • E.g. outward normal of (0,0,1) is stored as RGB(0.5,0.5,1), • The distinctive purple-blue of normal maps

  15. Parallax Mapping • Normal maps give the impression of bumpiness with correct lighting, but it’s only a lighting effect • The actual depth of bumpiness not shown • Parallax mappingtries to shows depth as well as bumpy lighting • Adjust the UVs used in the pixel shader • Distorts texture to give impression of depth • It’s an artificial effect ParallaxMapping Normal Mapping

  16. Idea behind Parallax Mapping • With parallax mapping we also need a height map: • A value for each texel indicating its height • Stored in the alpha channel of the normal map Height map Black low, white high

  17. Basics of Parallax Mapping • We have a “bumpy” texture, but we are applying it to a flat surface • This means that the texture coordinates we normally use are not the right ones: • We attempt to correct this by offsetting the UVs • Basic parallax mapping offsets them using the camera (eye) vector • Uses an approximation shown in diagrams

  18. Another Shader Example: Cell-Shading • Cell-Shading describes cartoon like rendering: • Black outlines and only a few colours • The outline can be created inside or outside of a shader • Here we draw a 2nd inside-out slightly larger black model • Other methods too (e.g. could look at polygon normals / camera) • Lighting is calculated in usual way • But colours are clamped to a fixed set of values • A 1D texture is used for this (see labs) • An example of non-photorealistic rendering

  19. Other Techniques • Many extensions to these techniques, e.g.: • Animated normal maps to create water effects • The parallax mapping methods can be taken further to create self shadowing surfaces with real depth • This is called steep parallax mapping • Newer approach: use tessellation to generate extra polygons to create the shape implied by the normal / height map • Many other techniques too, e.g.: • Different lighting models (e.g. for skin, metal, etc.) • Image processing, HDR lighting • Post-processing for blur, feedback, scene distortion, high dynamic range lighting etc. • And many more. Consider for a 3rd year project…

More Related